Beruflich Dokumente
Kultur Dokumente
LT 3209 Syntax
Prof. Matthias GERNER
Room B 7622
mgerner@cityu.edu.hk
http://www4.lt.cityu.edu.hk/~mgerner/
Q-8
Instructor:
Office:
e-mail:
Web:
Assignment Locker:
Outline of session:
1 NP MOVEMENT in Raising structures
2
3
EXTRAPOSITION
[NP ]
seems to me
[EC John
[I
to]
[VP be unhappy]]
[EC ----
[I
to]
[VP be unhappy]]
NP MOVEMENT
[NP John]
seems to me
This NP MOVEMENT in raising structures presupposes that John in (3) originates as the Subject of the
bracketed embedded S Complement, and is subsequently raised by application of NP MOVEMENT
to become the superficial Subject of the Main Clause. Similar to the previous movement analyses, we
have to show evidence for this presupposition to be right. First we have to show that John originates
as the Subject of the Exceptional Clause. Evidence emerges from facts related to the interpretation of
Reflexives. Reflexives always require a suitable antecedent from the same Clause, and not from a
different Clause. Thus, in (4) the italicized Reflexive himself can only refer to Fred but not to John.
(4) John considers [EC Fred to be too sure of himself].
Notice, however, that in a sentence such as
(5) John seems to me [EC ---- to be too sure of himself].
himself must be coreferential with John. How can we reconcile this with the above claim that a
reflexive form must always have a suitable antecedent of the same clause? In (5), John and himself
belong to two different clauses. The NP MOVEMENT analysis apparently can solve this question. It
supposes that John has originated from the position marked in (5) by ---. According to this
assumption, John and himself would belong to the same Clause at the D-structure level. Accordingly,
we have to revise our condition on reflexives:
(6) REFLEXIVES CONSTRAINT
A Reflexive can be construed with an NP which is contained in the same clause as the Reflexive at
some stage of derivation (either D-structure or S-structure level), and which is coreferential with it in
Person, Number, Gender etc.
The REFLEXIVE CONSTRAINT not only accounts for sentences such as (5), but also for structures of the
following type:
(7) John seems to himself [EC ---- to be invincible].
The REFLEXIVE CONSTRAINT predicts that the reflexive and its antecedent must belong to the same
clause at some stage of derivation. Even though John and himself do not belong to the same Clause at
the D-structure level (as was the case in (5)), they belong to the same Clause at the S-structure level.
For this reason, himself can construe an antecedent in (7).
The point that is important for our argumentation is that the REFLEXIVE CONSTRAINT supports the idea
of an NP MOVEMENT having taken place in sentences such as (5). There is more support for the idea
of an NP MOVEMENT in sentences such as (3) or (5) relating to Agreement facts. Generally speaking,
2
a Predicate Nominal (that is a noun phrase used as the Complement of a copular verb like be) agrees
in number with the Subject of its own Clause. Thus in
(8) They consider [S John to be a fool/*fools]
the Predicate Nominal (a fool/fools) agrees with the Subject John of its own S (the be-clause), not
with the Subject they of the main Clause. In the light of this observation, consider number agreement
of the following type:
(9) They seem to me [S ---- to be fools/*a fool]
Here, the Predicate nominal fools agrees with the italicized NP they, in spite of the fact that (as we
argued above) the two are contained in different Clauses at the surface level (i.e. S-structure level). If
we suppose that (9) is the result of an NP MOVEMENT of the following type, then we can account for
this apparent contradiction:
(10)
[NP ]
seem to me
[EC they [I to]
[VP be fools/*a fool]]
NP MOVEMENT
[NP They]
seem to me
[EC ----
[I
to]
[VP
be fools/*a fool]]
In (10), the subject they agrees with fools in the Complement Clause of seem at the D-structure level.
After the NP MOVEMENT the agreement properties of the deserted position are maintained even
though they has moved to the Subject position of the Main Clause.
Now, reviewing the two arguments about reflexives and about agreement, the question arises as
to whether the above sentences involving seem could be accounted for in other ways. Particularly, it
may seem plausible to view the seem-clauses as S-bar constituent at the D-structure level, similar to
the following sentences involving the predicate anxious:
(11) (a) I am anxious to finish the assignment
(b) I am anxious [S' C PRO to
finish the assignment]
(c) I am anxious [S' that I
should finish the assignment]
In this example, the S-structure sentence (11a) corresponds to the D-structure (11b), as it has the
same structure as the finite S-bar clause (11c). By symmetry, would it not be possible to propose
another interpretation for seem-clauses than the NP MOVEMENT analysis, notably the one of an S-bar
deep structure? Consider:
PRO
to
We have already seen above that the infinitival Complement of seem has the status of S (and not of
S-bar). It would follow from this that the infinitival Complement of seem cannot contain a PRO
Subject. For this reason the interpretation of raising-structures in terms of an S-bar constituent and a
PRO Subject cannot be successful.
For reference, we will call structures with an S-bar Complement Clause that has a covert C and a
PRO the control analysis, because PRO is controlled by the predicate of the Main Clause. The
corresponding predicate of the Main Clause will be called control predicate.
There are more differences between so-called control predicates like consider and raising
predicates like seem, especially semantic ones. For instance, raising predicates preserve semantic
equivalence in the process of passivization, control predicates do not. Consider:
(14) Raising predicate seem:
(a) The doctor seems to have examined John
(b) = John seems to have been examined by the doctor
(15) Control predicate anxious:
(a) The doctor is anxious to examine John
(b) John is anxious to be examined by the doctor
Furthermore, only control predicates appear to manifest selectional restrictions, whereas
raising predicates like seem do not appear to be sensible to selectional restrictions. Consider the
following examples:
(16) John/*the weather is anxious to change.
(17) John/the weather seems to have changed.
These arguments all point into the same direction. Raising predicates like seem cannot be
accounted for in the same way as control predicates can. This strengthens our basic assumption
4
that there is in fact an NP MOVEMENT in which the Subject of the Exceptional Clause is moved out
into Subject position of the Main Clause.
2 EXTRAPOSITION
We weve got to know three MOVEMENT transformations so far:
All three rules belong to what is known as substitution rules, because they substitute an empty
constituent by a filled one. In addition to these substitution rules, are there also other types of
MOVEMENT transformations?
It has been claimed that there is a second kind of MOVEMENT transformation known as adjunction
rule, and EXTRAPOSITION would be an example of adjunction rule. We can say that EXTRAPOSITION
applies to move a PP or S-bar contained within an NP (i.e. an adnominal PP or S-bar) to the end of
the S containing it. For example, it can move a PP Complement such as that italicized in the sentence
below out of the bracketed containing NP to the end of the whole S:
(18) (a) [A review of my latest book] has just appeared.
(b) [A review -----] has just appeared of my latest book.
Likewise, it can move a PP Adjunct such as that italicized in (19) out of the containing bracketed NP
to the end of the S:
(19) (a) [A gruesome figure with red cheeks] answered the door.
(b) [A gruesome figure ----] answered the door with red cheeks.
As we have studied in one of the Tutorials, it is also possible to extrapose an S-bar Complement
such as that italicized in (20), moving it out of the containing bracketed NP, to the end of the whole
S:
(20) (a) [A rumor that he is ill] has been circulating.
(b) [A rumor ----] has been circulating that he is ill.
Moreover, a clausal Adjunct (e.g. a Restrictive Relative Clause) can be extraposed out of its
containing NP in a similar fashion, as we see from:
[S [NP A review
[ PP of my latest book]]
As for previous transformation rules, the obvious question to ask is what evidence there is in support
of the EXTRAPOSITION analysis; and in particular, what evidence there is that the extraposed italicized
PP or S' in the (b) sentences in (18-21) above originate within the bracketed NP in the position they
occupy in the corresponding (a) sentences. One piece of evidence can be formulated in relation to
subcategorization facts
(23) (a) An unprecedented ATTACK on/*against/*at/*to the Press was launched by the President.
(b) A BAN on/*at/*to foreign imports has just been announced.
(c) Numerous CHANGES to/*at/*by the manuscript will be required.
(24) (a) An unprecedented ATTACK was launched by the President on/*against/*at/*to the Press.
(b) A BAN has just been announced on/*at/*to foreign imports.
(c) Numerous CHANGES will be required to/*at/*by the manuscript.
Subcategorization restrictions always hold between a Head and its sister constituents, it is clear that
the italicized PP in each case must originate as a sister Complement of the capitalized noun, within
an NP headed by the Noun in question.
A second argument in support of the extraction analysis is the familiar gap argument. If (as we
claim here) the italicized PP in sentences like (24) has indeed been moved out its underlying
containing Subject NP, then we should expect to find a gap after the capitalized head Noun of the
NP, which cannot be filled by another PP. And sure enough, this is indeed the case, as we see from
the ungrammaticality of sentences such as:
(25) (a) *An unprecedented ATTACK on the Press was launched by the President [on the Civil Right
Campaign].
(b) *A BAN [on foreign imports] has just been announced [on Japanese cars].
(c) *Numerous CHANGES [to the book] will be required [to the manuscript].
6
Under the EXTRAPOSITION analysis, the ungrammaticality of these sentences is simply to account for,
since both the italicized PP and the bracketed PP would have originated as sister Complements of the
capitalized head noun in violation of the subcategorization restriction that the Nouns in question
permit only one PP Complement of the relevant type.
We said above that EXTRACTION involves adjunction, then what is the constituent to which the
extraposed phrase is adjoined?
(26) (a) A review will appear in the Journal of Linguistics shortly of my latest book.
(b) *A review will appear of my latest book in the Journal of Linguistics shortly.
(c) *A review will appear in the Journal of Linguistics of my latest book shortly.
These sentences suggest that the extraposed constituent must follow all the constituents of VP
(including Adjunct Phrases). Thus, it would seem likely that extraposed constituents are adjoined
either to VP, or to S. But to which?
Lets consider first the case of extraposition out of an Object NP. Consider the following examples:
(27) (a) John will call [people who are from Boston] up.
(b) John will call [people ----] up who are from Boston.
(28) (a) John will call [people from Boston] up.
(b) John will call [people ----] up from Boston.
In both cases, the extraposed S-bar or PP is adjoined to the VP which dominates (=contains) the
Object NP of which it is Adjunct. The tree diagram structure of (27b) can be presented as follows:
(29)
S
NP
John
will
VP
VP
S'
call people up
This time, the extraposed S-bar constituent is adjoined to the S which dominates (=contains) the
Subject NP of which it is an adjunct. The tree diagram has the following outlook:
(31)
S
S
NP
Nobody
S'
VP
will
Combining both cases (29) and (31), we can pose the following ADJOINMENT RULE which regulates
the application of the EXTRAPOSITION movement rule.
(32) ADJOINMENT RULE FOR EXTRAPOSITION
An extraposed PP or S-bar is adjoined to the first major (i.e. phrasal or clausal, XP or S) constituent
dominating (containing) the NP out of which it is extraposed.
What we shall argue here first is that underlying the S(urface)-structure (40) is a more abstract level
of D-structure in which the NP [which car] originates within VP as the object of put. Secondly, we
shall also assume that the Modal will originates within I immediately following the NP subject [your
father]. In other words, we will claim that (40) derives from a more abstract D-structure of the
schematic form:
(41) [S' C [S Your father [I will] [VP put which car in the garage] ] ]
This derivation is achieved by the application of two movement rules. The first movement is the
familiar I MOVEMENT which moves the modal will out of I into the slot C of Complementizer. The
second movement rule, called WH MOVEMENT, moves the NP [which car] to a position within S-bar
in front of the preposed Modal will, as the leftmost constituent of S-bar. The operation of these two
movement rules can be represented in the following diagram:
(42) D-Structure: [S'
[C ] [S your father [I will] [VP put [NP which car] in the garage]]]
I MOVEMENT
WH MOVEMENT
There seems to be two different movement rules involved for two reasons. First, direct yes-no
questions show preposing of Modals (I MOVEMENT) without preposing of question-NPs
(WH-MOVEMENT):
(43) [S' Will [S your father put the car in the garage]]?
And secondly, indirect wh-questions show preposing of question-NPs (WH-MOVEMENT) without
preposing of Modals (I MOVEMENT):
(44) I dont know [S' which car [S your father will put ----- in the garage]]?
Examples (43) and (44) show that direct non-echo questions are derived through the application of
two rules, I MOVEMENT and WH MOVEMENT. However, the most controversial point in this
derivation is the claim that clause-initial wh-NPs which end up outside S after movement actually
originate inside S before movement. We will present syntactic as well as morphological evidence for
supporting this claim.
10